OFX interpretation query
A.J. Bonnema
gbonnema at xs4all.nl
Thu Dec 31 07:38:42 EST 2015
On 12/31/2015 03:18 AM, David Reiser wrote:
>> On Dec 30, 2015, at 6:26 PM, A.J. Bonnema <gbonnema at xs4all.nl> wrote:
>>
>>
>> It turns out that Gnucash processes both transactions as different posts correctly and in future imports ignores both transactions correctly.
>>
>> How is that possible? Can anyone explain how this mechanism works in Gnucash?
> Gnucash compares the FITID numbers of the current import batch against FITID numbers from prior batches. Once a transaction is accepted during the import process, GnuCash records internally (not viewable in the program interface) the FITID of each imported transaction.
>
> In your case, the conflicting FITID numbers were both in the same import batch, so comparison against existing FITIDs did not show any conflicts. Gnucash imported both transactions, and presumably you now have two transactions with identical FITID entries. Oops.
>
> You have found a hole in the importer that isn’t possible to reach with a OFX standards-compliant transaction list. One part of the specification states that a Financial Institution must ensure that all FITID’s are unique per account.
>
> Dave
> --
> Dave Reiser
> dbreiser at icloud.com
FITID is required to be unique, so the behavior for duplicate FITID is
to ignore the duplicates. This is perfectly acceptable.
Duplicate FITID in one import file shows a hole in the importer.
Of course, the current behavior happens to work out fine, and I would
not suggest this is high prio, unless the duplicates would cause
problems later in processing.
My suggestion would be to ignore the duplicate, but *not silently*. 2
identical *new* FITID's in the same import file should trigger a
warning, identifying the identical transactions. Or some kind of visual
warning, like not processing the transaction with the color red (Gnucash
does this sometimes, although it is not always clear why).
What do you think?
Kind regards, Guus.
More information about the gnucash-user
mailing list